Case Study Telecom Churn Analysis

• State: US state in which the customer resides

• Account length: The number of days that this account has been active

• Area code: The three-digit area code of the corresponding customer’s phone number

• International plan: Customer has an international calling plan: YES / NO

• Voice mail plan: Customer has a voice mail feature: YES / NO

• Number vmail messages: average number of voice mail messages per month

• Total day minutes: total number of calling minutes used during the day

• Total day calls: total number of calls placed during the day

• Total day charge: billed cost of daytime calls

• Total eve minutes: total number of calling minutes used during the evening

• Total eve calls: total number of calls placed during the evening

• Total eve charge: billed cost of evening time calls

• Total night minutes: total number of calling minutes used during the night

• Total night calls: the total number of calls placed during the night

• Total night charge: the billed cost of nighttime calls

• Total intl minutes: total number of international minutes

• Total intl calls: total number of international calls

• Total intl charge: billed cost for international calls

• Customer service calls: number of calls placed to Customer Service

• Churn: whether the customer left the service: True/False

Importing Dataset

Profile Report of dataset

Check basic information of dataset

14% Customer Left the service

Describe Dataset

Check Unique Value in dataset

Cleaning

Check the count of Null values in data set

There is no Null Value in Dataset

Churn Column Replace Bool Value to 0 & 1

EDA

Display Histogram of All numeric columns

Hear from Histo Grram, say that Account length, Total day minutes, Total day calls, Total day charge, Total eve minutes, Total eve calls, Total eve charge, Total night minutes, Total night calls, Total night charge, Total intl minutes, Total intl charge Columns are Normal Distribution

Check the churn column value count

Display count plot of Churn

Check the International plan column value count

Check the Voice mail plan column value count

Display the Pie Plot of International plan & Voice mail plan columns

Display the International plan & Voice mail plan count respect to Churn

Finds correlation between Independent and dependent attributes

We can go for Machine Learning

Feature Engineering

Apply Encoding in Categorical columns

One-Hot Encoding

Splitting the data in training and testing sets

Applying All Algorithms

RandomForestClassifier which shows an accuracy of nearly 94%

Here we see RandomForestRegressor gives us best score so we can use RandomForest Regressor algorithm

Algorithm selected: RandomForestClassifier

The next step consists of finding the combination of hyperparameters

Perform Hyper-Prameter tuning using RandomizedSearchCV

create list for all possible parameter

Performace of the model